Feasibility and analysis considerations of continuous community-based cough monitoring in low- and middle-income settings

Abstract Introduction Cough is a hallmark symptom of multiple respiratory conditions. Technological advances have made it possible to monitor cough using smartphone applications continuously and objectively. However, these technologies have not yet been evaluated for their feasibility in real-world research conditions. Methods We conducted 14-day continuous cough monitoring of presumptive TB patients enrolled as outpatients in a multi-center prospective tuberculosis diagnostics trial in Uganda, South Africa, Vietnam, the Philippines and India. Participants were given a smartphone with the Hyfe Research application installed, a carrier sling, and training on how to use both the phone and the cough application. We evaluated how often and how long participants recorded, and when coughs were detected. We also compared multiple cough frequency metrics for their performance in detecting changes in cough frequency. Results Of 693 participants who consented to continuous cough monitoring for 14 days, 568 (82.0%) completed at least one recording session successfully. Among these 568 participants, 421 (74.1%) recorded on at least 10 of the 14 days in the recording period for a median of 23.9 hours per day. We detected coughs in a diurnal pattern with a median of 1 cough per hour detected between 6am and 2pm local time and a median of 0 coughs for all other hours. We found median coughs per hour to be the cough frequency metric most robust to outlier values while still detecting changes in cough frequency during the recording period. Implementation challenges include the visibility of cough monitoring devices, nighttime recording and concurrent charging of devices. Conclusions Continuous cough monitoring using a smartphone application is feasible in low- and middle-income country settings.